CN104521232A

CN104521232A - Method and apparatus for coding/decoding image

Info

Publication number: CN104521232A
Application number: CN201380042182.2A
Authority: CN
Inventors: 金晖容; 林成昶; 李镇浩; 崔振秀; 金镇雄; 朴光勋; 金耿龙
Original assignee: Electronics and Telecommunications Research Institute ETRI; Industry Academic Cooperation Foundation of Kyung Hee University
Current assignee: Electronics and Telecommunications Research Institute ETRI; Industry Academic Cooperation Foundation of Kyung Hee University
Priority date: 2012-07-02
Filing date: 2013-07-02
Publication date: 2015-04-15
Also published as: US20150189289A1; JP7266515B2; JP2022116271A; US20180310006A1; CN115065823A; JP2015526013A; CN111629208B; JP2020048211A; US10187643B2; US20190356924A1; US10187644B2; JP2018137761A; JP2024038230A; CN115052157A; US20180310003A1; CN115052159A; US20180054621A1; CN112969073A; US20180310004A1; US10554982B2

Abstract

Disclosed are a method and an apparatus for coding/decoding an image. The method for decoding the image comprises the steps of: deriving a scale factor of a current block, depending on whether the current block is a conversion skip block; and scaling the current block on the basis of the scale factor, wherein the scale factor of the current block is derived on the basis of the position of a conversion coefficient inside the current block, and wherein the conversion skip block is the current block to which conversion is not applied and is specified on the basis of information indicating whether to apply reconversion to the current block.

Description

For the method and apparatus of encoding/decoding image

Technical field

The present invention relates to the coding/decoding of image, and more specifically, relate to the method and apparatus for scale transformation coefficient.

Background technology

There is the broadcast service of high definition (HD) resolution (1280x1024 or 1920x1080) in the whole nation and whole world expansion.Therefore, many user habits are in the video with high-resolution and high picture quality.Therefore, many mechanisms are just promoting the exploitation of image device of future generation.In addition, because exist for the ultrahigh resolution (UHD) of resolution and the growth interest of HDTV with higher than HDTV four times, so mobile image standardization organizes the demand becoming and recognize for having more high-resolution and the more compress technique of the image of high picture quality.In addition, exist for maintaining same frame quality and passing through the pressing needs that the compression efficiency higher than the compression efficiency H.264/AVC used now in HDTV, mobile phone and Blu-ray player also has the new standard of many advantages in frequency band or storage.

Nowadays, moving image expert group (MPEG) and Video Coding Experts group (VCEG) be the efficient video coding (HEVC) of combination with standard (namely, Video Codec of future generation), and object utilizes the compression efficiency of H.264/AVC twice to encode the image comprising UHD image.This can provide even has more low frequency and the more image of high picture quality and HD and UHD image than present image in 3D broadcast and mobile communications network.

Summary of the invention

[technical problem]

The invention provides can improve coding/decoding efficiency, for the method and apparatus of Code And Decode image.

The invention provides can improve coding/decoding efficiency, for the method and apparatus of scale transformation coefficient (or residual signals).

The invention provides can improve coding/decoding efficiency, for quantizing/method and apparatus of inverse quantization conversion skipped blocks.

[technical scheme]

According to an aspect of the present invention, a kind of picture decoding method is provided.This picture decoding method comprises: depend on whether current block is the zoom factor that conversion skipped blocks derives for current block, and perform convergent-divergent based on this zoom factor to current block.

The described zoom factor for current block is derived in place based on the conversion coefficient in current block, and this conversion skipped blocks is wherein also not to the block of current block application conversion, and specifies based on the information indicated whether to current block application inverse transformation.

Be used in described derivation in the step of the zoom factor of current block, if this current block is conversion skipped blocks, then can derive basic zoom factor, and the place of conversion coefficient no matter in current block.

This basic zoom factor can have specific values of zoom factor, and this specific values of zoom factor can be 16.

Whether this basic zoom factor can be depending on current block and uses quantization matrix and have different zoom factor values.

This basic zoom factor can be depending on current block and is luminance block or chrominance block and has different zoom factor values.

Indicate whether with signal transmission the mark using conversion jump algorithm in the picture comprising current block by parameter sets (PPS).

This basic zoom factor can comprise the information about the zoom factor for luminance signal and carrier chrominance signal.

Be used in described derivation in the step of the zoom factor of current block, if current block is conversion skipped blocks or current block do not use quantization matrix, then can derive basic zoom factor, and the place of no matter current block inner conversion coefficient.

Be used in described derivation in the step of the zoom factor of current block, if current block is not conversion skipped blocks, then quantization matrix can be used to derive zoom factor for current block based on the place of current block inner conversion coefficient.

According to a further aspect in the invention, a kind of image decoding apparatus is provided.This image decoding apparatus comprises: inverse quantization unit, for depending on whether current block is the zoom factor that conversion skipped blocks derives for current block, and performs convergent-divergent based on this zoom factor to current block.

The described zoom factor for current block can be derived based on the place of the conversion coefficient in current block, and this conversion skipped blocks can be wherein also not to the block of current block application conversion, and specifies based on the information indicated whether to current block application inverse transformation.

According to a further aspect in the invention, a kind of method for encoding images is provided.This method for encoding images comprises the following steps: depend on whether current block is the zoom factor that conversion skipped blocks derives for current block, and come to perform convergent-divergent to current block based on this zoom factor.

According to a further aspect in the invention, a kind of image encoding apparatus is provided.This image encoding apparatus comprises quantifying unit, for depending on whether current block is the zoom factor that conversion skipped blocks derives for current block, and comes to perform convergent-divergent to current block based on this zoom factor.

[advantageous effects]

Convert the block of jump algorithm to its application and to its existing piece of performing conversion/inversion process, there is different conversion coefficient characteristics, because not to performing conversion/inversion process to the block of its application conversion jump algorithm.That is, if to being applied to conversion skipped blocks to the scan method that it performs existing piece of application of conversion/inversion process, then coding/decoding efficiency can be reduced.Therefore, by applying zoom factor equally to conversion skipped blocks, and the place of conversion coefficient not in plumber block, Code And Decode efficiency can be improved.

Accompanying drawing explanation

Fig. 1 shows the block diagram of the structure of the image encoding apparatus that the embodiment of the present invention is applied to;

Fig. 2 shows the block diagram of the structure of the image decoding apparatus that the embodiment of the present invention is applied to;

Fig. 3 is the figure of the partitioned organization diagrammatically illustrating the image when coded image;

Fig. 4 shows the figure of the form of the PU that can comprise at CU;

Fig. 5 shows the figure of the form of the TU that can comprise at CU;

Fig. 6 be a diagram that the flow chart of the Zoom method for residual signals (or conversion coefficient) according to the embodiment of the present invention; With

Fig. 7 be a diagram that according to another embodiment of the present invention for the flow chart of the Zoom method of residual signals (or conversion coefficient).

Embodiment

Thereafter, example embodiment more of the present invention are described in detail with reference to accompanying drawing.In addition, when describing the embodiment of this specification, by omitting the detailed description of known function and formation, if it is fuzzy to think that it makes main idea of the present invention unnecessary.

In this specification, when thinking that an element is connected with another element or couples, this might mean that a described element can directly be connected with another element described or couple, or third element can connect or couple between these two elements during.In addition, in this specification, when think comprise particular element time, this might mean that the element do not got rid of except this particular element, and add ons can be included in the scope of embodiments of the invention or technical spirit of the present invention.

The term of such as first and second can be used to describe various element, but these elements are not limited to these terms.These terms are used only an element and another element to be distinguished.Such as, the first element can be called as the second element, and does not depart from the scope of the present invention.Equally, the second element can be called as the first element.

In addition, the cell described in embodiments of the invention is independently shown, to indicate difference and feature functionality, and it is not intended each cell and formed by an independent hardware or a software.Namely, for convenience of description, arrange and comprise these cells, and in these cells at least two can form a cell, or an element can be divided into multiple cell, and described multiple divided cell can n-back test.Wherein integrated component embodiment or be also included within the scope of this invention from its embodiment being separated some elements, unless they depart from marrow of the present invention.

In addition, in the present invention, some elements are not the necessary elements for performing necessary function, and can be only for improvement of the selectable unit of performance.The element that the present invention only can use the necessary element for realizing marrow of the present invention instead of be used for only improving performance realizes, and only comprises necessary element and the structure not comprising the selectable unit for only improving function is included within the scope of this invention.

First, in order to improve the facility of description and help understanding of the present invention, the term used in concise and to the point this specification of description.

Unit means Image Coding or decoding unit.In other words, when coding or decoded picture, coding or decoding unit represent the division unit of an image when carrying out son to image and dividing and encode or decode.This unit also can be called as block, macro block (MB), coding unit (CU), predicting unit (PU), converter unit (TU), encoding block (CB), prediction block (PB) or transform block (TB).A unit can be divided into less subelement.

Block represents the MxN array of sample.M and N has positive integer value.Block can mean the array of 2-D form jointly.

Converter unit (TU) is the elementary cell when performing coding/decoding to residual signals, such as transformation series transformation of variable, inverse transformation, quantification, inverse quantization and coding/decoding.A TU can be partitioned multiple less TU.Here, if there is residual signals in block form, then residual signals can be called as residual block.

Quantization matrix means at the matrix quantized or use in inverse quantization process, to improve subjectivity or the objective picture-quality of image.Quantization matrix is also referred to as convergent-divergent list.

Quantization matrix can be divided into default matrix, non-default matrix and flat (flat) matrix.Default matrix can mean particular quantization matrix predetermined in the encoder and the decoder.Non-default matrix can not be predetermined in encoder, but may imply that the quantization matrix that user transmits or receives.Flat matrix may imply that wherein all elements has the matrix of identical value.

Convergent-divergent represents the process of conversion coefficient level being multiplied by the factor.As the result of this process, generate conversion coefficient.Convergent-divergent is also referred to as inverse quantization.

Conversion coefficient represents the coefficient value generated after performing conversion.In this specification, after the quantification obtained by quantizing to conversion coefficient application, conversion coefficient level is also referred to as conversion coefficient.

Quantization parameter represents in the value quantized and in inverse quantization, scale transformation coefficient levels uses.Here, quantization parameter can be the value being mapped to quantization step size.

Parameter set corresponds to the information about the header in the structure in bit stream.Parameter set has the implication of common specified sequence parameter set, parameter sets and auto-adaptive parameter collection.

Fig. 1 shows the block diagram of the structure of the image encoding apparatus that the embodiment of the present invention is applied to.

With reference to figure 1, image encoding apparatus 100 comprises motion estimation module 111, motion compensating module 112, intra-framed prediction module 120, switch 115, subtracter 125, conversion module 130, quantization modules 140, entropy code module 150, inverse quantization module 160, inverse transform module 170, adder 175, filter module 180 and reference picture buffer 190.

Image encoding apparatus 100 can perform coding according to frame mode or inter-frame mode to input picture, and output bit flow.When frame mode, switch 115 can be switched to frame mode.When inter-frame mode, switch 115 can be switched to inter-frame mode.Infra-frame prediction means frame intra-prediction, and inter prediction means between frame.Image encoding apparatus 100 can generate the prediction block of the input block being used for input picture, and the difference then between coding input block and prediction block.Here, input picture can mean raw frames.

When frame mode, intra-framed prediction module 120 performs spatial prediction by the value of the pixel using the coded block adjacent with current block, carrys out generation forecast block.

When inter-frame mode, motion estimation module 111, by search in the reference picture of storage in reference picture buffer 190 in motion prediction process and the region of input block optimum Match, obtains motion vector.Motion compensating module 112 performs motion compensation by the reference picture using motion vector and store in reference picture buffer 190, carrys out generation forecast block.Here, motion vector is two dimension (2-D) vector used in inter prediction, and motion vector can indicate the skew between picture and reference picture wanting present coding/decoding.

Subtracter 125 based on the difference between input block and the prediction block of generation, can generate residual block.

Conversion module 130 can perform conversion to residual block, and according to the block output transform coefficient after conversion.In addition, quantization modules 140, by quantizing the conversion coefficient received according to quantization parameter, carrys out the coefficient after output quantization.

The value that entropy code module 150 can calculate based on quantization modules 140, the encoded parameter values etc. calculated in the encoding process, perform entropy code according to probability distribution to code element, and carry out output bit flow according to the code element after entropy code.If application entropy code, then can reduce the size of the bit stream for the code element that will encode, because pass through to the symbol allocation peanut bit with the high frequency of occurrences and to the symbol allocation big figure bit with the low frequency of occurrences, represent code element.Therefore, the compression performance of Image Coding is improved by entropy code.Entropy code module 150 can use such as index Columbus, the coding method of context-adaptive variable length code (CAVLC) and context adaptive binary arithmetic coding (CABAC) is used for entropy code.

Perform inter prediction encoding (that is, predictive coding between frame) according to the image encoding apparatus 100 of the embodiment of Fig. 1, and picture encoded thus needs and stores decoded, to be used as reference picture.Therefore, the coefficient after quantification is by inverse quantization module 160 inverse quantization and by inverse transform module 170 inverse transformation.Coefficient after inverse quantization and inverse transformation adds prediction block to by adder 175, generates reconstructed blocks thus.

Reconstructed blocks is through (experiences) filter module 180.Filter module 180 can one or more in reconstructed blocks or reconstruction picture application deblocking filter, sample self adaptation skew (SAO) and auto-adaptive loop filter (ALF).Filter module 180 also can be called as filter in self adaptation ring.Deblocking filter can remove the block distortion generated at the boundary of block.SAO can add suitable deviant to compensate encoding error to pixel value.ALF based on by comparing reconstruction picture and raw frames and the value obtained, can perform filtering.The reconstructed blocks of wave filter module 180 can be stored in reference picture buffer 190.

Fig. 2 shows the block diagram of the structure of the image decoding apparatus that the embodiment of the present invention is applied to.

With reference to figure 2, image decoding apparatus 200 comprises entropy decoder module 210, inverse quantization module 220, inverse transform module 230, intra-framed prediction module 240, motion compensating module 250, filter module 260 and reference picture buffer 270.

Image decoding apparatus 200 can receive the bit stream exported from encoder, comes to perform decoding to bit stream, and export reconstructed image (that is, the image of reconstruct) according to frame mode or inter-frame mode.When frame mode, switch can be switched to frame mode.When inter-frame mode, switch can be switched to inter-frame mode.

Image decoding apparatus 200 can obtain the residual block of reconstruct from the bit stream received, generation forecast block, and by adding the residual block of reconstruct to prediction block, generates reconstructed blocks (that is, recovery block).

Entropy decoder module 210, by performing entropy decoding according to probability distribution to the bit stream received, generates the code element of the code element comprising the coefficient form after having quantification.

If application entropy decoding method, then can reduce the size of the bit stream for each code element, because pass through to the symbol allocation peanut bit with the high frequency of occurrences and to the symbol allocation big figure bit with the low frequency of occurrences, represent code element.

Coefficient after quantification is by inverse quantization module 220 inverse quantization and by inverse transform module 230 inverse transformation.As the result of the inverse quantization/inverse transformation to the coefficient after quantification, the residual block of reconstruct can be generated.

When frame mode, intra-framed prediction module 240 performs spatial prediction by using the value of the pixel of the coded block around current block, carrys out generation forecast block.When inter-frame mode, motion compensating module 250 performs motion compensation by using the reference picture stored in motion vector and reference picture buffer 270, carrys out generation forecast block.

Together with residual block is added to prediction block by adder 255.The block device module 260 be after filtering added.Filter module 260 can at least one in the block of reconstruct or the picture application deblocking filter of reconstruct, SAO and ALF.Filter module 260 exports reconstructed image (that is, the image of reconstruct).Reconstructed image can be stored in reference picture buffer 270 also can be used for inter prediction.

Fig. 3 is the figure of the partitioned organization diagrammatically illustrating the image when coded image.

In efficient video coding (HEVC), in coding unit, perform coding so that efficient zoned image.

With reference to figure 3, in HEVC, image 300 is sequential partition in maximum coding unit (being called LCU thereafter), and determines partitioned organization based on LCU.Partitioned organization means the distribution of the coding unit (being called CU thereafter) for the image in efficient coding LCU310.Whether can will be partitioned four CU (its each width dimensions and height dimension reduce half from a CU) based on a CU, determine this distribution.Equally, the CU of subregion can be four CU by recursive partitioning, and its each width dimensions and height dimension reduce half from the CU of subregion.

Here, the subregion of CU can be performed at the most (up to) desired depth by recurrence.Information about the degree of depth is the information of the size of instruction CU, and stores the information about the degree of depth of each CU.Such as, the degree of depth of LCU can be 0, and the degree of depth of minimum code unit (SCU) can be predetermined depth capacity.Here, LCU is the CU with above-mentioned maximum CU size, and SCU is the CU with minimum CU size.

As long as perform the subregion of width dimensions and height dimension half from LCU 310, the degree of depth of CU just increases by 1.Also the CU of subregion is not performed to it, for each degree of depth, there is 2N × 2N size, and to its CU performing subregion from the CU with 2N × 2N size be partitioned its each there are four CU of N × N size.As long as the size that the degree of depth increases by 1, N just reduces half.

With reference to figure 3, the size with the LCU of minimum-depth 0 can be 64 × 64 pixels, and the size with the SCU of depth capacity 3 can be 8 × 8 pixels.Here, the LCU with 64 × 64 pixels can represent by the degree of depth 0, and the CU with 32 × 32 pixels can represent by the degree of depth 1, and the CU with 16 × 16 pixels can represent by the degree of depth 2, and the SCU with 8 × 8 pixels can represent by the degree of depth 3.

In addition, whether the partition information of the information be partitioned by 1 bit being used for each CU is represented about specific CU.This partition information can be included in all CU except SCU.Such as, if CU is not partitioned, then can partition holding information 0.If CU is partitioned, then can partition holding information 1.

Therebetween, predicting unit (PU) (or prediction block (PB)) can be comprised (namely from the CU of LCU subregion, elementary cell for predicting) and converter unit (TU) (or transform block (TB)) (that is, the elementary cell for converting).

Fig. 4 shows the figure of the form of the PU that can comprise at CU.

From among the CU of LCU subregion, the CU that do not rezone is partitioned one or more PU.The behavior oneself is also referred to as subregion.Predicting unit (being called PU thereafter) be to its perform prediction elementary cell, and dancing mode, inter-frame mode and frame mode any one in encode.PU can depend on that each pattern carrys out subregion according to various forms.

With reference to figure 4, when dancing mode, in CU, the 2Nx2N pattern 410 with CU with same size can be supported without the need to subregion.

When inter-frame mode, the form of 8 kinds of subregions can be supported in CU, such as, 2Nx2N pattern 410,2NxN pattern 415, Nx2N pattern 420, NxN pattern 425,2NxnU pattern 430,2NxnD pattern 435, nLx2N pattern 440 and nRx2N pattern 445.

When frame mode, 2Nx2N pattern 410 and NxN pattern 425 can be supported in CU.

Fig. 5 shows the figure of the form of the TU that can comprise at CU.

Converter unit (being called TU thereafter) is the elementary cell used in order to the spatial alternation in CU and quantification/inverse quantization (convergent-divergent) process.TU can have rectangle or square form.From among the CU of LCU subregion, the CU that do not rezone can be partitioned one or more TU.

Here, the partitioned organization of TU can be quad-tree structure.Such as, as shown in Figure 5, it is one or more that CU 510 can depend on that quad-tree structure is partitioned, and forms the TU with various sizes thus.

Therebetween, in HEVC, in H.264/AVC, inter prediction (being called infra-frame prediction thereafter) coding can be performed.Here, by deriving the intra prediction mode (or prediction direction) for current block from the adjacent block be positioned near current block, coding is performed.

As mentioned above, the predicted picture of the signal obtained by performing prediction based on intra prediction mode can have the difference with original image.The residual image with the difference between predicted picture and original image can stand entropy code after experience frequency domain transform and quantization.Here, in order to increase the code efficiency of frequency domain conversion, can depend on the size of block, and the integer transform of intra prediction mode, discrete cosine transform (DCT), discrete sine transform (DST) or DCT/DST are depended in selectivity and adaptability application.

In addition, in order to increase file and picture in such as PowerPoint or speech image, code efficiency in screen content, can use and convert jump algorithm.

If use conversion jump algorithm, then the residual image (or residual block) of encoder to the difference had between predicted picture and original image directly quantizes, and without the need to frequency conversion process, and entropy code is performed to residual block.In addition, decoder performs entropy decoding to residual block, and generates the residual block of reconstruct by performing inverse quantization (convergent-divergent) to the block after entropy code.Therefore, frequency translation/inversion process is skipped to the block of its application conversion jump algorithm.

In the process of quantification/inverse quantization, the place of the conversion coefficient in block can be depended on and differently apply zoom factor, to improve the main body image quality of image.On the contrary, there is the place of the conversion coefficient when execution quantification/inverse quantization not in plumber block and apply the method for zoom factor in the same manner.Signal whether apply the method by the sequence parameter set (SPS) of bit stream or parameter sets (PPS).

As the embodiment of this process, the convergent-divergent process for conversion coefficient can be performed as follows.

for the convergent-divergent process of conversion coefficient

In this case, input is as follows.

The width of-Current Transform block; NW

The height of-Current Transform block; NH

-there is element c _ijthe array of conversion coefficient; (nWxnH) array d

-for the luminance signal of current block and the index of carrier chrominance signal; CIdx

If cIdx is 0, then this means luminance signal.If cIdx be 1 or cIdx be 2, then this means carrier chrominance signal.In addition, if cIdx is 1, then this means the Cb in carrier chrominance signal.If cIdx is 2, then this means the Cr in carrier chrominance signal.

-quantization parameter; QP

In this case, output is as follows.

The array of conversion coefficient after-convergent-divergent: (nWxnH) array d _ij

Derived parameter " log2TrSize " is carried out by log2TrSize=(Log2 (nW)+Log2 (nH)) >>1.Depend on cIdx and differently derived parameter displacement.If cIx is 0 (when luminance signal), then from " shift=BitDepth _y+ log2TrSize – 5 " carry out derived parameter displacement.If cIx is not 0 (when carrier chrominance signal), then from " shift=BitDepth _c+ log2TrSize – 5 " carry out derived parameter displacement.Here, BitDepth _yand BitDepth _cmean the number (such as, 8 bits) of the bit of the sample for present image.

Array " levelScale [] " for zooming parameter is identical with following equation 1.

[equation 1]

[k]={ 40,45,51,57,64,72} is k=0..5 wherein for levelScale

The conversion coefficient after convergent-divergent is calculated by following process.

First, zoom factor m is derived by following process _ij.

If-scaling_list_enable_flag is 0, then as following equation 2, derive zoom factor m _ij.

[equation 2]

m _ij＝16

If-scaling_list_enable_flag is not 0, then as following equation 3, derive zoom factor m _ij.

[equation 3]

m _ij＝ScalingFactor[SizeID][RefMatrixID][trafoType][i*nW+j]

In equation 3, the size according to transform block derives SizeID by following form 1, and derives RefMatrixID and trafoType respectively from following equation 4 and equation 5.In addition, in equation 4, by sequence parameter set (SPS) or the parameter sets (PPS) of bit stream, scaling_list_pred_matrix_id_delta is transmitted with signal.

[equation 4]

RefMatrixID＝MatrixID-scaling_list_pred_matrix_id_delta

[equation 5]

trafoType＝((nW＝＝nH)？0:((nW>nH)？1:2))

Form 1 shows the example of the SizeID value of the size according to transform block

[form 1]

The size of quantization matrix	SizeID
		4x4	0
8x8(16x4,4x16)	1

16x16(32x8,8x32)	2
		32x32	3

Next, the conversion coefficient d after convergent-divergent is derived from following equation 6 _ij.

[equation 6]

d _ij＝Clip3(-32768,32767,((c _ij*m _ij*levelScale[qP％6]＜＜(qP/6))+(1＜＜(shift-1)))＞＞shift)

Therebetween, not to performing frequency conversion process to the block (being called thereafter conversion skipped blocks) of its application conversion jump algorithm as mentioned above.Therefore, different conversion coefficient characteristic can be had to existing piece that it performs frequency conversion process with conversion skipped blocks.That is, if by not alternatively being applied to conversion skipped blocks to the Zoom method that it performs existing piece of application of frequency conversion process, then can code efficiency be reduced.

Therefore, the invention provides by consider wherein block be conversion skipped blocks situation, perform the method for convergent-divergent.

If use quantization matrix (default matrix and non-default matrix) to improve the subjective image quality of image in the encoder and the decoder, then can depend on that the zoom factor of deriving from quantization matrix is differently applied in the place of the conversion coefficient in block.In the method, when converting this block, by use wherein by the energy compression of residual block to block upper left (namely, low frequency region) characteristic, the more insensitive high-frequency region of human eye (instead of low frequency region of human eye sensitivity) is performed there is the quantification of relatively largeization step size.According to the method, when coded image, the subjective image quality in the region of human eye sensitivity can be improved.

But, if application conversion jump algorithm, then residual block is not compressed towards the low frequency region in residual block, because do not perform frequency domain conversion/inverse transformation to residual block.In this case, if be applied in the quantification/quantification method used in existing frequency domain, then the shortcoming existed is that the distortion in image or block becomes serious.Therefore, if use quantization matrix in image, then exist for can the demand of convergent-divergent (quantification/inverse quantization) method of distortion in the block (that is, converting skipped blocks) it not being performed to frequency domain conversion/inverse transformation, in reduction image or block.Such as, exist not to the method for conversion skipped blocks application quantization matrix.In the method, the basic zoom factor of application can be equal to, and the place of conversion coefficient not in plumber block.

[embodiment 1] for the place of the conversion coefficient in not plumber block to the equivalent application of conversion skipped blocks the method and apparatus of zoom factor

Fig. 6 be a diagram that the flow chart of the Zoom method for residual signals (or conversion coefficient) according to the embodiment of the present invention.

The Zoom method of Fig. 6 can perform in the decoding device of the encoding device of Fig. 1 or Fig. 2.More specifically, the Zoom method of Fig. 6 can perform in the quantifying unit of Fig. 1 or 2 or inverse quantization unit.In the embodiment in fig 6, although the Zoom method of Fig. 6 is illustrated as and performs in encoding device for convenience of description, the Zoom method of Fig. 6 can be equal to application in decoding device.

With reference to figure 6, can depend on whether current block is conversion skipped blocks, and derive the zoom factor m of the application when performing convergent-divergent (quantizing or inverse quantization) to the residual signals (or conversion coefficient) in current block _ij.

In step S600, encoding device determines whether current block is conversion skipped blocks.

Whether can be the information converting skipped blocks based on instruction current block, determine whether current block is conversion skipped blocks.Such as, whether indicate current block to be the information of conversion skipped blocks can be mark " transSkipFlag ".By to the information and executing entropy decoding about conversion skipped blocks in bit stream, derive the value of mark " transSkipFlag ".If current block is conversion skipped blocks, then indicate that the value of " transSkipFlag " can be 1.If current block is not conversion skipped blocks, then indicate that the value of " transSkipFlag " can be 0.

If be conversion skipped blocks (such as, the value of mark " transSkipFlag " is 1) as determination result determination current block, then encoding device derives zoom factor m in step S610 _ijand the no matter place of residual signals (or conversion coefficient) in current block.

Here, as shown in Figure 6, zoom factor m _ijspecific basic values of zoom factor T can be set to.Such as, this specific basic values of zoom factor T can be 16.

If be not convert skipped blocks (such as determination result determination current block, the value of mark " transSkipFlag " is 0), then encoding device derives zoom factor m in step S620 based on the place of residual signals in current block (or conversion coefficient) _ij.

Here, quantization matrix can be used to depend on and zoom factor m is differently set for residual signals (or conversion coefficient) in current block _ij.As shown in Figure 6, zoom factor m can be derived as in equation 7 _ij.

[equation 7]

m _ij＝ScalingFactor[SizeID][RefMatrixID][trafoType][i*nW+j]

In equation 7, ScalingFactor is the array wherein storing zoom factor.SizeID can be the value of size of instruction current block (transform block or quantization matrix), and the value of SizeID can depend on that the size of current block (transform block) derives as in form 1 above.RefMatrixID and trafoType can derive respectively from equation 8 and equation 9.NW is the width of current block.

[equation 8]

RefMatrixID＝MatrixID-scaling_list_pred_matrix_id_delta

In equation 8, the value of MatrixID can mean the type of the quantization matrix according to predictive mode and color component.Such as, the value of MatrixID can derive as in form 2 below.By the sequence parameter set (SPS) in bit stream or parameter sets (PPS), transmit scaling_list_pred_matrix_id_delta with signal.

[equation 9]

trafoType＝((nW＝＝nH)？0:((nW>nH)？1:2))

In equation 9, nW means the width of current block, and nH means the height of current block.

Form 2 shows the MatrixID value according to predictive mode and color component.

[form 2]

The Zoom method of Fig. 7 can perform in the decoding device of the encoding device of Fig. 1 or Fig. 2.More specifically, the Zoom method of Fig. 7 can perform in the quantifying unit of Fig. 1 or 2 or inverse quantization unit.In the embodiment of Fig. 7, although the Zoom method of Fig. 7 is illustrated as and performs in encoding device for convenience of description, the Zoom method of Fig. 7 can be equal to application in decoding device.

With reference to figure 7, can depend on whether current block is conversion skipped blocks and whether uses quantization matrix, and derive the zoom factor m of the application when the residual signals (or conversion coefficient) to current block performs convergent-divergent (quantizing or inverse quantization) _ij.

In step S700, encoding device determines whether current block uses whether quantization matrix and current block are conversion skipped blocks.

Whether can use the information of quantization matrix based on instruction current block, determine whether current block uses quantization matrix.Such as, current block is indicated whether to use the information of quantization matrix can be mark " scaling_list_enable_flag ".By decoding about the information and executing entropy of the use of quantization matrix in bit stream, derive the value of mark " scaling_list_enable_flag ".If current block uses quantization matrix, then indicate that the value of " scaling_list_enable_flag " can be 1.If current block does not use quantization matrix, then indicate that the value of " scaling_list_enable_flag " can be 0.

In addition, whether can be the information converting skipped blocks based on instruction current block, determine whether current block is conversion skipped blocks.Such as, whether indicate current block to be the information of conversion skipped blocks can be mark " transSkipFlag ".By to the information and executing entropy decoding about conversion skipped blocks in bit stream, derive the value of mark " transSkipFlag ".If current block is conversion skipped blocks, then indicate that the value of " transSkipFlag " can be 1.If current block is not conversion skipped blocks, then indicate that the value of " transSkipFlag " can be 0.

If be that conversion skipped blocks or current block do not use quantization matrix (such as determination result determination current block, transSkipFlag==1 or scaling_list_enable_flag==0), then encoding device derives zoom factor m in step S710 _ij, and the no matter place of residual signals (or conversion coefficient) in current block.

As shown in Figure 7, zoom factor m _ijspecific basic values of zoom factor T can be set to.Such as, this specific basic values of zoom factor T can be 16.

If be not conversion skipped blocks as determination result determination current block and current block use quantization matrix, then encoding device derives zoom factor m in step S720 based on the place of residual signals in current block (or conversion coefficient) _ij.

Zoom factor m _ijquantization matrix can be used to depend on, and the place of residual signals in current block (or conversion coefficient) is differently arranged, and can derive as in the equation shown in the step S720 of Fig. 7.The zoom factor m that equation shown in step S720 is derived _ijstep S620 with reference to figure 6 is described, and the descriptions thereof are omitted.

As described by Fig. 6 and 7, if current block (namely, the object block will encoded or decode now) be conversion skipped blocks, then (namely the zoom factor with particular value T is applied to current block, conversion skipped blocks), and the no matter place of coefficient (or signal) in current block.Here, can be depending on each coding parameter to corresponding blocks application according to the value of the zoom factor of the embodiment of the present invention and differently arrange.

Such as, can depend on the value indicating whether the parameter (such as, scaling_list_enable_flag) using quantization matrix, arranging as follows will to the value of the zoom factor of corresponding blocks application.

If-using quantization matrix (such as, scaling_list_enable_flag==1), then basic values of zoom factor is set to " T1 " (m _ij=T1)

If-not using quantization matrix (such as, scaling_list_enable_flag==0), then basic values of zoom factor is set to " T2 " (m _ij=T2)

T1 and/or T2 value can be determined by encoder and use signal transmission, or can use predetermined T1 and/or T2 value.If transmit T1 and/or T2 value by bit stream signal, then decoder obtains T1 and/or T2 value by resolving this bit stream.

For another example, can based on the value of the information (such as, color component index cIdx) of the color characteristics about the signal can derived for corresponding blocks, arranging as follows will to the value of the zoom factor of corresponding blocks application.Color component index cIdx depends on its value and indicates luminance signal (that is, Y-signal) or carrier chrominance signal (that is, Cb signal or Cr signal).

-example 1: whether the signal depending on corresponding blocks is luminance signal, and basic values of zoom factor is set to " Ty " or " Tc ".Such as, if the signal of corresponding blocks is luminance signal, then basic values of zoom factor is set to " Ty ".If if the signal of corresponding blocks is not luminance signal (that is, carrier chrominance signal), then basic values of zoom factor is set to " Tc ".

-example 2: each color component according to corresponding blocks arranges basic values of zoom factor.Such as, if the color component of corresponding blocks is luminance signal (that is, Y-signal), then basic values of zoom factor is set to " Ty ".If carrier chrominance signal is Cb signal, then basic values of zoom factor is set to " Tcb ".If carrier chrominance signal is Cr signal, then basic values of zoom factor is set to " Tcr ".

Here, Ty, Tc, Tcb and/or Tcr value can be determined by encoder and use signal transmission, or can use predetermined Ty, Tc, Tcb and/or Tcr value.If transmit Ty, Tc, Tcb and/or Tcr value by bit stream signal, then decoder obtains Ty, Tc, Tcb and/or Tcr value by resolving this bit stream.

According to the embodiment of the present invention for depending on that coding odd number determines that the place of basic zoom factor can independence or Combination application, but need always to identical conversion skipped blocks application same zoom factor values, and the place of coefficient (or signal) no matter in this block (that is, the object block encoded or decode).

The convergent-divergent process for conversion coefficient that embodiments of the invention have been merged into can be performed as follows.

for the convergent-divergent process of conversion coefficient

In this case, input is as follows.

The width of-Current Transform block; NW

The height of-Current Transform block; NH

-there is element c _ijthe array of conversion coefficient; (nWxnH) array d

-indicate whether to the information of Current Transform block application conversion jump algorithm

-for the luminance signal of Current Transform block and the index of carrier chrominance signal; CIdx

-quantization parameter; QP

In this case, output is as follows.

Derived parameter " log2TrSize " is carried out by " log2TrSize=(Log2 (nW)+Log2 (nH)) >>1 ".The differently derived parameter displacement according to cIdx.If cIx was 0 (that is, when luminance signal), then from " shift=BitDepth _y+ log2TrSize – 5 " carry out derived parameter displacement.If cIx was not 0 (that is, when carrier chrominance signal), then from " shift=BitDepth _c+ log2TrSize – 5 " carry out derived parameter displacement.Here, BitDepth _yand BitDepth _cmean the number (such as, 8 bits) of the bit of the sample for present image.

Array " levelScale [] " for zooming parameter is identical with equation 10.

[equation 10]

[k]={ 40,45,51,57,64,72} is k=0..5 wherein for levelScale

By conversion coefficient after following process calculating convergent-divergent.

First, zoom factor m is derived by following process _ij.

If-scaling_list_enable_flag be 0 or Current Transform block be conversion skipped blocks, then as following equation 11, derive zoom factor m _ij.

[equation 11]

m _ij＝16

-if not, then as following equation 12, derive zoom factor m _ij.

[equation 12]

m _ij＝ScalingFactor[SizeID][RefMatrixID][trafoType][i*nW+j]

In equation 12, the size according to transform block derives SizeID by above form 1, and derives RefMatrixID and trafoType respectively from following equation 13 and equation 14.In addition, in equation 13, by the sequence parameter set (SPS) of bit stream, scaling_list_pred_matrix_id_delta is transmitted with signal.

[equation 13]

RefMatrixID＝MatrixID-scaling_list_pred_matrix_id_delta

[equation 14]

trafoType＝((nW＝＝nH)？0:((nW>nH)？1:2))

Next, conversion coefficient d after deriving convergent-divergent from equation 15 below _ij.

[equation 15]

Therebetween, as mentioned above inversion process is performed to the conversion coefficient by convergent-divergent process convergent-divergent.Here, do not perform inversion process, but only the Current Transform block that conversion jump algorithm has been applied to is performed with shift-down oepration process.

If 1. the cIdx of current block is 0 (when luminance signal), then shift (displacement)=13-BitDepth _y.If the cIdx of current block is not 0 (when carrier chrominance signal), then shift=13-BitDepth _c.

2. the following array r arranged for residual block _ij(i=0 ... (nW)-1, j=0.. (nH)-1).

If displacement is greater than 0, then r _ij=(d _ij+ (1<< (shift-1))) >>shift.If displacement is not more than 0, then r _ij=(d _ij<< (-shift).

Here, d _ijthe array of conversion coefficient after convergent-divergent, and r _ijmean the array of the residual block obtained by performing inverse transformation to conversion coefficient after convergent-divergent.

As the embodiment that the inversion process of conversion coefficient after convergent-divergent has been merged into, the conversion process for conversion coefficient after convergent-divergent can be performed as follows.

for the conversion process of conversion coefficient after convergent-divergent

In this case, input is as follows.

The width of-Current Transform block; NW

The height of-Current Transform block; NH

-there is element d _ijthe array of conversion coefficient; (nWxnH) array d

-quantization parameter; QP

In this case, output is as follows.

-the array of residual block that obtains by performing inverse transformation to conversion coefficient after convergent-divergent; (nWxnH) array r

If be intra prediction mode for the coding mode " PredMode " of current block, the value of Log2 (nW*nH) is 4, and the value of cIdx is 0, then depend on that the intra prediction mode of luminance signal obtains parameter " horizTrType " and " vertTrType " by following form 3.If not, then parameter " horizTrType " and " vertTrType " are set to 0.

Form 3 shows the example of the value of parameter " horizTrType " according to intra prediction mode and " vertTrType ".

[form 3]

IntraPredMode	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
																			vertTrType	1	0	0	0	0	0	0	0	0	0	0	1	1	1	1	1	1	1
horizTrType	1	0	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1

IntraPredMode	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34
																		vertTrType	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1	1
horizTrType	1	1	1	1	1	1	1	1	0	0	0	0	0	0	0	0	0

The residual signals being used for current block is obtained according to following order.

First, if applied the conversion jump algorithm for current block, then performed following process.

If 1. cIdx is 0, then shift=13 – BitDepth _y.If cIdx is not 0, then shift=13 – BitDepth _c.

2. the following array r arranged for residual block _ij(i=0.. (nW)-1, j=0.. (nH)-1).

If-displacement is greater than 0, then r _ij=(d _ij+ (1<< (shift-1))) >>shift.If displacement is not more than 0, then r _ij=(d _ij<< (-shift).

If also do not apply the conversion jump algorithm for current block, then perform following process.

1. the value of operation parameter " horizTrType " and " vertTrType " is come to perform inversion process to conversion coefficient after convergent-divergent.First, receive the size (nW of current block, nH), for the array " (nWxnH) array d " of conversion coefficient after convergent-divergent and parameter " horizTrType ", and perform 1 dimension inversion by level and bring output array " (nWxnH) array e ".

2. following, receiving array " (nWxnH) array e ", and as in equation 16, derive array " (nWxnH) array g ".

[equation 16]

g _ij＝Clip3(-32768,32767,(e _ij+64)＞＞7)

3. following, receive the size (nW, nH) of current block, array " (nWxnH) array g " and parameter " vertTrType ", and level performs 1 dimension inverse transformation.

4. following, depend on that cIdx arranges the array " (nWxnH) array r " for residual block as in equation 17.

[equation 17]

r _ij＝(f _ij+(1＜＜(shift-1)))＞＞shift

In equation 17, when cIdx is 0, shift=20 – BitDepth _y.If not, then shift=20 – BitDepth _c.BitDepth means the number (such as, 8 bits) of the bit of the sample for present image.

By performing above-mentioned convergent-divergent process to conversion coefficient and performing above-mentioned conversion process to conversion coefficient after convergent-divergent, the residual block of reconstruct can be generated.In addition, by adding to the residual block of reconstruct the prediction block generated by infra-frame prediction or inter prediction, and the block of reconstruct is generated.Here, the block of reconstruct can be to the block of its application loop filter or also not to the block of its application loop filter.

Thereafter, the invention provides and depend on that whether current block is conversion skipped blocks and the method for basic zoom factor that derives with signal transmission.

According to embodiments of the invention, depend on whether current block is conversion skipped blocks and the basic zoom factor of deriving uses signal transmission by sequence parameter set (SPS).

Form 4 shows transmitting about the example of the SPS grammer of the information of basic zoom factor with signal according to the embodiment of the present invention.

[form 4]

Reference table 4, transform_skip_enabled_flag indicates whether to use conversion jump algorithm in current sequence.

If use conversion jump algorithm in current sequence, then transmit flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 with signal.Each value can encode (se (v)) according to the form with plus or minus symbol.Or each value can encode (ue (v)) according to the form with 0 and plus sign.

Flat_scale_factor_y_minus16 means the zoom factor for luminance signal.Such as, if the value of flat_scale_factor_y_minus16 is 0, then the zoom factor for luminance signal has value 16, wherein adds 16 to 0.

Flat_scale_factor_cb_minus16 means the zoom factor for carrier chrominance signal Cb.Flat_scale_factor_cr_minus16 means the zoom factor for carrier chrominance signal Cr.

The zoom factor for luminance signal or carrier chrominance signal can be derived as equation 18 to 20.

Here, the storage of basic zoom factor FlatScalingFactor [cIdx] is used for the zoom factor of luminance signal and carrier chrominance signal.Such as, if color component index cIdx is 0, then basic zoom factor can indicate brightness (Y) signal.If color component index cIdx is 1, then basic zoom factor can indicate Cb carrier chrominance signal.If color component index cIdx is 2, then basic zoom factor can indicate Cr carrier chrominance signal.In addition, the value of basic zoom factor " FlatScalingFactor [cIdx] " can have the scope of particular value.Such as, 8 bit signals can have from-15 to the value of " 255-16 ".

The basic zoom factor for luminance signal can be derived as in equation 18.

[equation 18]

FlatScalingFactor[0]＝16+(transform_skip_enabled_flag＝＝1)？:flat_scale_factor_y_minus16:0

The basic zoom factor for Cb carrier chrominance signal can be derived as in equation 19.

[equation 19]

FlatScalingFactor[1]＝16+(transform_skip_enabled_flag＝＝1)？:flat_scale_factor_cb_minus16:0

The basic zoom factor for Cr carrier chrominance signal can be derived as in equation 20.

[equation 20]

FlatScalingFactor[2]＝16+(transform_skip_enabled_flag＝＝1)？:flat_scale_factor_cr_minus16:0

By by the method for basic zoom factor that derives is merged in convergent-divergent process according to the depending on whether current block is conversion skipped blocks with signal transmission of the embodiment of the present invention, the convergent-divergent process for conversion coefficient can be performed as follows.

for the convergent-divergent process of conversion coefficient

In this case, input is as follows.

The width of-Current Transform block; NW

The height of-Current Transform block; NH

-there is element c _ijthe array of conversion coefficient; (nWxnH) array d

-indicate whether to the information of Current Transform block application conversion jump algorithm; TransSkipFlag

If the value of transSkipFlag is 1, then its instruction is to current block application conversion jump algorithm.If the value of transSkipFlag is 0, then its instruction is not also to current block application conversion jump algorithm.

-quantization parameter; QP

In this case, output is as follows.

Derived parameter log2TrSize is carried out by log2TrSize=(Log2 (nW)+Log2 (nH)) >>1.Depend on cIdx and differently derived parameter displacement.If cIx is 0 (when luminance signal), then from " shift=BitDepth _y+ log2TrSize – 5 " carry out derived parameter displacement.If cIx was not 0 (that is, when carrier chrominance signal), then from " shift=BitDepth _c+ log2TrSize – 5 " carry out derived parameter displacement.Here, BitDepth _yand BitDepth _cmean the number (such as, 8 bits) of the bit of the sample for present image.

Array " levelScale [] " for zooming parameter is identical with equation 21.

[equation 21]

[k]={ 40,45,51,57,64,72} is k=0..5 wherein for levelScale

Conversion coefficient after convergent-divergent is calculated by following process.

First, zoom factor m is derived by following process _ij.

If-scaling_list_enable_flag is 0, then as in following equation 22, derive zoom factor m _ij.

[equation 22]

m _ij＝(transSkipFlag＝＝1)？FlatScaleFactor[cIdx]:16

-if not (if that is, scaling_list_enable_flag is 1), then as in following equation 23, derive zoom factor m _ij.

[equation 23]

m _ij＝(transSkipFlag＝＝1)？FlatScaleFactor[cIdx]:ScalingFactor[SizeID][ReMatrixID][trafoType][i*nW+j]

In equation 23, depend on that the size of block derives SizeID by above form 1.RefMatrixID and trafoType is derived respectively from equation 24 below and equation 25.In equation 24, transmit scaling_list_pred_matrix_id_delta by the sequence parameter set (SPS) of bit stream with signal.

[equation 24]

RefMatrixID＝MatrixID-scaling_list_pred_matrix_id_delta

[equation 25]

trafoType＝((nW＝＝nH)？0:((nW>nH)？1:2))

Next, conversion coefficient dij after deriving convergent-divergent from equation 26.

[equation 26]

Therebetween, except above-mentioned SPS, by parameter sets (PPS) or section headers " SliceHeader ", transmit with signal and depend on that whether current block is conversion skipped blocks and the basic zoom factor of deriving according to the embodiment of the present invention.In addition, in CU unit or TU unit, basic zoom factor can be transmitted with signal.

The value flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 that are transmitted by above-mentioned SPS signal can be upgraded and use in PPS (or SliceHeader, CU or TU).

Form 5 shows and transmits the example about the PPS grammer of the information of basic zoom factor with signal according to another embodiment of the present invention.

[form 5]

Whether reference table 5, transform_skip_enabled_flag indicates conversion jump algorithm will use in current picture.If use conversion jump algorithm, then transmit pps_flat_scaling_factor_present_flag with signal.

Such as, if the value of pps_flat_scaling_factor_present_flag is 0, then use flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 of applying to above-mentioned SPS as the zoom factor for converting skipped blocks.If the value of pps_flat_scaling_factor_present_flag is 1, then transmit respective value with signal, to upgrade the value of flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and the flat_scale_factor_cr_minus16 applied in above-mentioned SPS.

The zoom factor of the conversion skipped blocks for current picture is used as by the value " flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 " that signal transmits.Here, continue to use these values, until they do not change again.Or, only can apply these values to current picture, and can be applied in next picture the values of zoom factor used in SPS.

Here, each in value " flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 " can encode (se (v)) according to the form with plus or minus symbol.Or each in these values can encode (ue (v)) according to the form with 0 and plus sign.

Different value signal delivery value " flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 " can be utilized according to each luminance signal and each carrier chrominance signal.Such as, can use value " flat_scale_factor_y_minus16 " come with the zoom factor of signal transmission for luminance signal, use value " flat_scale_factor_cb_minus16 " can come with the zoom factor of signal transmission for Cb carrier chrominance signal, and can use value " flat_scale_factor_cr_minus16 " come with the zoom factor of signal transmission for Cr carrier chrominance signal.Or, flat_scale_factor_y_minus16 can be used to come with the zoom factor of signal transmission for luminance signal, and flat_scale_factor_cb_cr_minus16 can be used to come with the zoom factor of signal transmission for carrier chrominance signal.Or, a value " flat_scale_factor_y_cb_cr_minus16 " can be used to transmit zoom factor for luminance signal and carrier chrominance signal with signal.

As mentioned above, can upgrade and use in SliceHeader (or CU or TU) by the value " flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 " that signal transmits in SPS or PPS.

Form 6 shows and transmits the example about section headers " SliceHeader " grammer of the information of basic zoom factor with signal according to another embodiment of the present invention.

[form 6]

Reference table 6, transform_skip_enabled_flag indicates whether to use conversion jump algorithm in current clip.If by use conversion jump algorithm, then transmit the value of flat_scaling_factor_override_flag with signal.

Such as, if the value of flat_scaling_factor_override_flag is 0, be then used in flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and flat_scale_factor_cr_minus16 of applying in above-mentioned SPS or PPS as the zoom factor for converting skipped blocks.If the value of flat_scaling_factor_override_flag is 1, then transmit respective value with signal, to upgrade the value of flat_scale_factor_y_minus16, flat_scale_factor_cb_minus16 and the flat_scale_factor_cr_minus16 applied in above-mentioned SPS or PPS.

Use value " flat_scale_factor_y_delta, flat_scale_factor_cb_delta and flat_scale_factor_cr_delta " is as the zoom factor of the conversion skipped blocks for current clip.

Here, each of value " flat_scale_factor_y_delta, flat_scale_factor_cb_delta and flat_scale_factor_cr_delta " can encode (se (v)) according to the form with plus or minus symbol.Or each value can encode (ue (v)) according to the form with 0 and plus sign.

Different value signal delivery value " flat_scale_factor_y_delta, flat_scale_factor_cb_delta and flat_scale_factor_cr_delta " can be utilized according to each luminance signal and each carrier chrominance signal.Such as, can come with the zoom factor of signal transmission for luminance signal by use value flat_scale_factor_y_delta, can come with the zoom factor of signal transmission for Cb carrier chrominance signal by use value flat_scale_factor_cb_delta, and can come with the zoom factor of signal transmission for Cr carrier chrominance signal by use value flat_scale_factor_cr_delta.Or, flat_scale_factor_y_delta can be used to come with the zoom factor of signal transmission for luminance signal, and flat_scale_factor_cb_cr_delta can be used to come with the zoom factor of signal transmission for carrier chrominance signal.Or, a value flat_scale_factor_y_cb_cr_delta can be used to transmit zoom factor for luminance signal and carrier chrominance signal with signal.

Can utilize and as in equation 27 to 29, derive basic zoom factor by the value " flat_scale_factor_y_delta, flat_scale_factor_cb_delta and flat_scale_factor_cr_delta " that signal transmits as mentioned above.

Here, basic zoom factor " FlatScalingFactor [cIdx] " stores the zoom factor being used for luminance signal and carrier chrominance signal.Such as, if color component index cIdx is 0, then basic zoom factor can indicate brightness (Y) signal.If color component index cIdx is 1, then basic zoom factor can indicate Cb carrier chrominance signal.If color component index cIdx is 2, then basic zoom factor can indicate Cr carrier chrominance signal.In addition, the value of basic zoom factor " FlatScalingFactor [cIdx] " can indicate the scope of particular value.Such as, 8 bit signals can have from-15 to the scope of " 255-16 ".

Can use flat_scale_factor_y_delta as in equation 27, derive basic zoom factor for luminance signal.

[equation 27]

FlatScalingFactor[0]＝16+(transform_skip_enabled_flag＝＝1)？:(flat_scale_factor_y_minus16+flat_scale_factor_y_delta):0

Can use flat_scale_factor_cb_delta as in equation 28, derive basic zoom factor for Cb carrier chrominance signal.

[equation 28]

FlatScalingFactor[1]＝16+(transform_skip_enabled_flag＝＝1)？:(flat_scale_factor_cb_minus16+flat_scale_factor_cb_delta):0

Can use flat_scale_factor_cr_delta as in equation 29, derive basic zoom factor for Cr carrier chrominance signal.

[equation 29]

FlatScalingFactor[2]＝16+(transform_skip_enabled_flag＝＝1)？:(flat_scale_factor_cr_minus16+flat_scale_factor_cr_delta):0

Therebetween, above-described embodiment can depend on the degree of depth of the size of block, the degree of depth of CU or TU and have different ranges of application.Determine that the parameter (such as, about the size of block or the information of the degree of depth) of range of application can be arranged by encoder, make this parameter have predetermined value or can be set to according to distribution map or level and have predetermined value.When encoder in the bitstream write parameters value time, decoder can obtain this value from bit stream and use this value.

If range of application depends on the degree of depth of CU and different, then following three kinds of methods can be applied to above-described embodiment as illustrated in form 7.Method A is only applied to has certain depth or the higher degree of depth, and method B is only applied to has certain depth or the lower degree of depth, and method C is only applied to certain depth.

Form 7 shows for depending on that the degree of depth of CU (or TU) determines the example of the method for the scope wherein applying method of the present invention.In form 7, mark " O " means corresponding good application corresponding method to CU (or TU), and marks " X " and mean not to the corresponding good application corresponding method of CU (or TU).

[form 7]

The degree of depth of the CU (or TU) of the scope of instruction application	Method A	Method B	Method C
				0	×	○	×
1	×	○	×
				2	○	○	○
3	○	×	×
				4 or higher	○	×	×

Reference table 7, if the degree of depth of CU (or TU) is 2, then can apply all method A, method B and method C to embodiments of the invention.

If not to all good application embodiments of the invention of CU (or TU), then it can use certain indicators (such as, mark) indicate, or by using the value of the degree of depth of the CU of instruction range of application, representing by the value of signal transmission larger than the maximum of the degree of depth of CU 1.

In addition, for depending on that the above-mentioned degree of depth of CU (or TU) determines that the method for the scope wherein applying method of the present invention can depend on the size of luminance block and chrominance block and differently apply, and luminance picture and chromatic diagram picture can be depended on and differently apply.

Form 8 shows to schematically show and depends on that the size of luminance block and chrominance block is to determine the example of the combination of the method for range of application.

[form 8]

When method " G 1 " in the method listed in form 8, if the size of luminance block is 8 (8x8,8x4,2x8 etc.) and the size of chrominance block is 4 (4x4,4x2,2x4), then embodiments of the invention 1 (G1-embodiment 1) can be applied to luminance signal and carrier chrominance signal and horizontal signal and vertical signal.

In above demonstration system, although the flow chart based on the form of succession of steps or block describes these methods, but the invention is not restricted to the order of these steps, and some steps can perform according to the order different from other steps, or can perform with other steps simultaneously.In addition, it will be understood by those skilled in the art that the step shown in flow chart is not exclusiveness, and these steps can comprise additional step, or the one or more steps in flow chart can be deleted, and not affect scope of the present invention.

More than describing is only the example of technical spirit of the present invention, and those skilled in the art can change according to various mode and revise the present invention, and does not depart from intrinsic propesties of the present invention.Therefore, the disclosed embodiments should not be interpreted as limiting technical spirit of the present invention, but should be interpreted as illustrating technical spirit of the present invention.The scope of technical spirit of the present invention is not subject to the restriction of these embodiments, and should explain scope of the present invention based on claims.Therefore, the present invention should be interpreted as covering all modifications or modification of drawing from implication and the scope of claims and equivalence thereof.

Claims

1. a picture decoding method, comprises step:

Depend on whether current block is the zoom factor that conversion skipped blocks derives for current block; With

Based on this zoom factor, convergent-divergent is performed to current block,

Wherein based on the place of the conversion coefficient in current block, derive the described zoom factor for current block, and

This conversion skipped blocks is wherein also not to the block of current block application conversion, and specifies based on the information indicated whether to current block application inverse transformation.

2. picture decoding method according to claim 1, the step of zoom factor that wherein said derivation is used for current block comprises: if this current block is conversion skipped blocks, then derive basic zoom factor, and the place of conversion coefficient no matter in current block.

3. picture decoding method according to claim 2, wherein:

This basic zoom factor has specific values of zoom factor, and

This specific values of zoom factor is 16.

4. picture decoding method according to claim 2, wherein this basic zoom factor depends on whether current block uses quantization matrix and have different zoom factor values.

5. picture decoding method according to claim 2, wherein this basic zoom factor depends on that current block is luminance block or chrominance block and has different zoom factor values.

6. picture decoding method according to claim 1, wherein indicates whether with signal transmission the mark using conversion jump algorithm in the picture comprising current block by parameter sets (PPS).

7. picture decoding method according to claim 6, wherein this basic zoom factor comprises the information about the zoom factor for luminance signal and carrier chrominance signal.

8. picture decoding method according to claim 1, the step that wherein said derivation is used for the zoom factor of current block comprises: if current block is conversion skipped blocks or current block do not use quantization matrix, then derive basic zoom factor, and the place of no matter current block inner conversion coefficient.

9. picture decoding method according to claim 1, the step of zoom factor that wherein said derivation is used for current block comprises: if current block is not conversion skipped blocks, then use quantization matrix to derive zoom factor for current block based on the place of current block inner conversion coefficient.

10. an image decoding apparatus, comprising:

Inverse quantization unit, for depending on whether current block is the zoom factor that conversion skipped blocks derives for current block, and performs convergent-divergent based on this zoom factor to current block,

The described zoom factor for current block is derived in place wherein based on the conversion coefficient in current block, and

11. 1 kinds of method for encoding images, comprise the following steps:

Come to perform convergent-divergent to current block based on this zoom factor,

12. method for encoding images according to claim 11, the step of zoom factor that wherein said derivation is used for current block comprises: if this current block is conversion skipped blocks, then derive basic zoom factor, and the place of conversion coefficient no matter in current block.

13. method for encoding images according to claim 12, wherein:

This basic zoom factor has specific values of zoom factor, and

This specific values of zoom factor is 16.

14. method for encoding images according to claim 12, wherein this basic zoom factor depends on whether current block uses quantization matrix and have different zoom factor values.

15. method for encoding images according to claim 12, wherein this basic zoom factor depends on that current block is luminance block or chrominance block and has different zoom factor values.

16. method for encoding images according to claim 11, wherein indicate whether with signal transmission the mark using conversion jump algorithm in the picture comprising current block by parameter sets (PPS).

17. method for encoding images according to claim 16, wherein this basic zoom factor comprises the information about the zoom factor for luminance signal and carrier chrominance signal.

18. method for encoding images according to claim 11, the step that wherein said derivation is used for the zoom factor of current block comprises: if current block is conversion skipped blocks or current block do not use quantization matrix, then derive basic zoom factor, and the place of no matter current block inner conversion coefficient.

19. method for encoding images according to claim 11, the step of zoom factor that wherein said derivation is used for current block comprises: if current block is not conversion skipped blocks, then use quantization matrix to derive zoom factor for current block based on the place of current block inner conversion coefficient.

20. 1 kinds of image encoding apparatus, comprising:

Quantifying unit, for depending on whether current block is the zoom factor that conversion skipped blocks derives for current block, and comes to perform convergent-divergent to current block based on this zoom factor,